基于双模驱动的耙吸式挖泥船生产率预测

Productivity Prediction of Trailing Suction Dredger Based on Dual Mode Drive

为解决耙吸式挖泥船疏浚预测系统预测的生产率不准确和严重依赖泥浆浓度数据的问题,提出一种在缺少泥浆浓度参数的情况下,利用模型叠加概化方法准确预测挖泥船生产率的数据挖掘方法。剔除异常施工数据,并对数据进行归一化处理和对数平滑变换。采用Spearman秩相关系数法进行特征提取。介绍5种机器学习模型,即套索模型、弹性网络模型、梯度提升决策树模型、极限梯度提升模型和轻型梯度提升模型,在这5种模型的基础上,应用一个叠层综合模型。结果表明,该模型的拟合优度R2为0.9177,其精度高于其他算法,优化效果明显。

In order to solve the problems that the dredging prediction system provides inaccurate productivity prediction and relies heavily on the mud concentration data,a data mining method to accurately predict the productivity of dredgers by using the model superposition generalization method under the condition of lacking mud concentration parameters is proposed.The abnormal construction data are eliminated,and then the data are normalized and log smooth transformed.Spearman rank correlation coefficient method is used for feature extraction.Five machine learning models are introduced,including lasso model,elastic network model,gradient lifting decision tree model,limit gradient lifting model and light gradient lifting model.On the basis of these five models,a stack synthesis model is applied.The results show that the goodness of fit R2 of the model is 0.9177,which is higher than the accuracy of other algorithms,and the optimization effect is obvious.